Research On Parameter Estimation Of Software Reliability Model Based On Hybrid Wolf Group Algorithm

Today,Computer technology is growing faster and faster,software plays a very important role in all walks of life,such as: medical,financial,transportation and so on.Therefore,the quality of these software must be improved,and software reliability is one of the important attributes for evaluating software quality.The evaluation of software reliability is mainly realized by models.There are hundreds of models currently existed,but most of the models are nonlinear,and it is difficult to estimate their parameters,so choose the appropriate model for reliability.The assessment is very important.In recent years,many scholars have proposed a new idea for different software reliability models,which is to apply the swarm intelligence optimization algorithm to the software reliability model parameter estimation.Maximum likelihood and least squares are the two most commonly used estimation methods,but these two methods are easy to damage the constraints of reliability model parameter estimation in the process of optimization,and reduce the accuracy of understanding.Based on the above problems,this paper uses WPA and PSO algorithms and a mixture of the two to estimate and predict the software reliability model parameters.The parameters of the GO model are estimated and predicted using five sets of classical software failure numbers.Then the software reliability model of the wolf group algorithm and the particle swarm algorithm are studied in depth,and the fitness function is constructed by the maximum likelihood method.The wrong solution is removed during the operation of the algorithm.The software reliability model parameters are estimated using a wolf group algorithm,a particle swarm algorithm,and a mixture of the two.The wolf group algorithm has strong global optimization ability,the algorithm has fast convergence speed,and the optimization strategy is diverse,but the algorithm is relatively complicated.The particle swarm algorithm has a simple structure and fast convergence,but it is easy to fall into precocity and the accuracy of the solution is not high.Based on the advantages and disadvantages of the two algorithms,the hybrid algorithm is applied to the parameter estimation of the model.The experimental results show that the accuracy of the algorithm is greatly improved in both parameter estimation and prediction.Especially when the algorithm runs dozens of times,the solution of the hybrid algorithm running dozens of times is more compared with a single algorithm.Close to the actual value,further illustrates the accuracy of the hybrid algorithm estimation parameters.